| Because of the security and repeatability of imaging technology, in order to better grasp the emergence and spread of disease, people have been extensive use of imaging technology. Magnetic resonance imaging (MRI) can clearly reflect the form of the soft tissues, and especially distinguish the white matter (WM), gray matter (GM) and cerebrospinal fluid (CSF) in the brain tissue. At present, extraction and shape analysis of the human cerebral cortex is an important way for quantitative analysis of neuroscience and kinds of nerve diseases (especially Alzheimer's disease). However, the extraction of cortical surface faces a huge challenge at home and abroad, for example, most of the methods need to provide very near to the initial conditions and manual intervention, extracted surface can not keep the completeness on the topology and a heavy computation requirement. These defects lead to difficultly calculate surface curvature, seriously affect the shape analysis and increase the calculation error of cortical volume and average thickness.I proposes a extraction of the inner and outer surfaces of the cerebral cortex from3D MR images through two-phase fast matching method. In the first phase,3D arrival-time field can be generated from the MR images through the fast marching method, from which the inner and outer cortical surface can be extracted. Due to the noise of image date, the part of volume effect, image artifacts, non-uniformity and other factors, the isosurface connectivity is poor. In the second phase, in order to solve the connectivity problem, the implicit surfaces (isosurfaces) are defined by3D arrival-time field, on which the fast marching method can performed to remove some 'island'from generated isosurfaces.A new approach to compute the cortical volume is given. By comparison of our calculated result and the expert manual classification of MRI gray matter volume. The experimental results show that the relative error is1.35-1.42%. |
PDF Full Text Request